Ecology, 85(6), 2004, pp. 1627±1634 ᭧ 2004 by the Ecological Society of America

MATERNAL NEST-SITE CHOICE AND OFFSPRING FITNESS IN A TROPICAL ( MAIRII, )

G. P. BROWN AND R. SHINE1 Biological Sciences A08, University of Sydney, NSW 2006 Australia

Abstract. Do reproducing female adaptively manipulate phenotypic traits of their offspring by selecting appropriate nest sites? Evidence to support this hypothesis is indirect, mostly involving the distinctive characteristics of used (vs. available) nest sites, and the fact that physical conditions during egg incubation can modify hatchling phenotypic traits that plausibly might in¯uence ®tness. Such data fall well short of demonstrating that nesting females actively select from among potential sites based on cues that predict ®tness- determining phenotypic modi®cations of their offspring. We provide such data from ex- perimental studies on a small oviparous snake (the keelback, ) from the wet-dry tropics of Australia. When presented with a choice of alternative nesting sites, egg-laying females selected more moist substrates for egg deposition. Incubation on wetter substrates signi®cantly increased body size at hatching, a trait under strong positive selection in this population (based on mark±recapture studies of free-ranging hatchlings). Remark- ably, the hydric conditions experienced by an egg in the ®rst few hours after it was laid substantially affected phenotypic traits (notably, muscular strength) of the hatchling that emerged from that egg 10 weeks later. Thus, our data provide empirical support for the hypothesis that nesting female reptiles manipulate the phenotypic traits of their offspring through nest-site selection, in ways that enhance offspring ®tness. Key words: Australia; egg incubation; keelback snake; nest-site selection; offspring ®tness; phe- notypic plasticity; ; Tropidonophis mairii.

INTRODUCTION natural selection. In the extreme case in which all phe- Within most populations, and especially in notypic variation within a cohort of offspring is due to sexually reproducing , a cohort of neonates dis- environmental, not genetic, factors (as in a clone of plays substantial phenotypic variation. Charles Dar- parthenogens), even intense selection on offspring win's greatest insight was that such variation provides traits will not generate any longer term (evolutionary) an opportunity for natural selection to modify the dis- response because of the absence of genetic variation tribution of ®tness-relevant traits, thereby increasing at the critical loci. This is not to say, however, that the the frequency of characteristics that enhance an organ- system cannot evolve (Via et al. 1995). If speci®c en- ism's probability of surviving and reproducing. The vironmental conditions result in the development of degree to which such selective forces result in evolu- ``®tter'' phenotypic traits, we expect selection for any tionary change within the population depends, how- behaviors that expose the offspring to such environ- ever, upon the proximate mechanisms that generate that ments at the appropriate time within their life history. phenotypic variation. In the simplest case, the pheno- Often, this will involve early development, generally typic variation is engendered entirely by genetic var- the most critical phase because even small deviations iation, so that ®tness differentials will directly modify in early embryos can later cascade through into major underlying gene frequencies. Unfortunately, the reality phenotypic modi®cations as development unfolds (Tan- is more complex. One important complication is the ing 1952, Albon et al. 1983, Henry and Ulijaszek fact that phenotypic variance within a cohort is the 1996). Thus, one of the most important proximate in- result of environmental in¯uences as well as genetic ¯uences on offspring phenotypes may be the conditions factors. Indeed, a high proportion of the quanti®able that eggs experience in natural nests. Minor shifts in variation in many traits is induced by environmental traits such as nest temperatures and water potentials factors, not by genes (Bull 1987, Sultan 1987, War- can have major impacts on phenotypic traits of hatch- kentin 1995, Scheiner and Callahan 1999). lings such as sex, size, shape, color, locomotor ability, This sensitivity of phenotypic traits to environmental and behavior (Burger et al. 1987, Deeming and Fer- conditions has strong implications for the operation of guson 1991, Rhen and Lang 1995). Hence, one major route by which natural selection can modify offspring Manuscript received 13 February 2003; revised 6 October traits in such a system is via genes that control the nest- 2003; accepted 7 October 2003; ®nal version received 31 October site selection behavior of reproducing females (Bull et 2003. Corresponding Editor: D. K. Skelly. 1 Author to whom correspondence should be addressed. al. 1982, Bull 1983, Packard and Packard 1988, Pack- E-mail: [email protected] ard 1991). 1627 1628 G. P. BROWN AND R. SHINE Ecology, Vol. 85, No. 6

Many studies of this topic have been based on rep- Cogger 2000). This species is abundant through many tiles. Laboratory-based experimental work has shown tropical and subtropical areas within the Australasian a high degree of phenotypic plasticity in hatchling rep- region, especially around bodies of water (O'Shea tiles as a result of the conditions experienced during 1991, Cogger 2000), and feeds primarily upon frogs embryogenesis (e.g., Joanen et al. 1987, Ji and Brana (Shine 1991). Keelbacks have been extensively studied 1999, Warner and Andrews 2000, Webb et al. 2001). on the Adelaide River ¯oodplain 60 km east of Darwin Quantitative comparisons of natural nests with avail- in the Australian wet-dry tropics (Webb et al. 2001, able nest sites often reveal substantial differences, pro- Brown and Shine 2002, Brown et al. 2002). Ambient viding strong (albeit indirect) evidence that reproduc- temperatures at this site are high year-round (mean ing females actively select particular types of sites for monthly temperature 27.0ЊC), but precipitation is high- nesting (Muth 1980, Packard and Packard 1988). Last- ly seasonal. More than 78% of the 1394 mm mean ly, ®eld studies suggest that a neonate's phenotype may annual rainfall comes from monsoonal downpours in¯uence its probability of survival (e.g., Fox 1975, within the relatively brief (four-month) ``wet season'' Ferguson and Fox 1984, Arnold and Bennett 1988, Jay- (December±March). Thus, much of the ¯oodplain is ne and Bennett 1990). In combination, these three kinds inundated (and soils in surrounding higher areas are of studies support the hypothesis that mothers may be saturated) during the wet season, but soil moisture lev- able to manipulate the phenotypic traits of their off- els fall gradually over the course of the next several spring by exploiting norms of reaction of reptilian em- months (Shine and Brown 2002). bryogenesis in relation to physical conditions during Keelbacks on the Adelaide River ¯oodplain nest over egg incubation (Beuchat 1986, 1988, Beuchat and Ell- an eight-month period (April±November), with some ner 1987, Shine and Harlow 1996, Arnold and Peterson females producing multiple clutches within the same 2002). That is, some component of selection on off- year (Brown and Shine 2002). Eggs are laid in rela- spring phenotypes in such systems is mediated not tively shallow (Ͻ20 cm) burrows in the ¯oodplain soil through differential ®tness of alleles that determine (Shine and Brown 2002). Thus, clutches laid at dif- speci®c offspring traits, but instead through selection ferent times of year experience different hydric con- on genes in females for nest-site selection criteria. ditions during incubation. Experimental studies have Unfortunately, the available evidence remains indi- shown that the phenotypic traits of hatchling keelbacks rect. One problem is that maternal choice among al- are in¯uenced signi®cantly by the physical conditions ternative potential nesting sites has been inferred most- that the eggs experience during incubation. In partic- ly from nonrandom attributes of natural nests rather ular, hatchling phenotypes are affected by the magni- than from experimental studies that manipulate nest tude of diel variation in nest temperatures (Webb et al. availability. Correlational evidence cannot identify the 2001) and by the water potential of the incubation me- actual cue(s) used by nesting females, because so many dium (Shine and Brown 2002). of these cues (e.g., temperature/moisture/soil depth/ type and size of cover object) covary in nature. Ideally, Experimental methods we need to manipulate the array of available nest sites As is true for many species, natural nests of keel- with respect to a speci®c attribute (such as temperature backs are dif®cult to locate (Shine and Brown 2002). or moisture) relevant to the reaction norms of the off- To examine the cues used by nesting females, we cap- spring. Only then can we make a strong link between tured females that had migrated to the wall of Fogg nest-site selection and its impact on offspring pheno- Dam to lay their eggs (Brown and Shine 2002). After types to discern (1) whether or not the female actively the females were measured and weighed, we offered selects among potential nest sites, and if so, (2) which them a choice of potential nest sites differing in hydric cues does she use. We can then (3) examine the impact conditions. Females oviposited 2±18 days after collec- of variation in that cue on hatchling phenotypes, to tion (mean ϭ 7.7 days). During this period, they were construct a direct causal link between female behavior kept individually in clear plastic cages (40 ϫ 30 ϫ 20 and offspring traits. Lastly, we need ®eld data to (4) cm) with a water bowl and four potential nest sites. explore the ways in which this environmentally in- These were circular black plastic bowls 10 cm in di- duced variation translates into ®tness differentials. Our ameter and 4.5 cm high, covered by a plastic lid with studies on the ecology of in tropical Australia a 3 cm diameter central opening. The bowls contained provide such data. vermiculite moistened with water to produce substrates ranging from very dry (0.25 g water per 1 g vermiculite, METHODS ϭ 25%) to very wet (6 g water per 1 g vermiculite, ϭ 600%). Bowls were reweighed daily and water was Study species and area added to bring them back to their original mass (i.e., The keelback, Tropidonophis mairii, is a small, non- to counteract evaporative water loss, generally Ͻ0.1 g/ (adult size Ͻ80 cm snout±vent length d). In 37 initial trials, we offered the snakes a choice [SVL hereafter]) belonging to the natricine lineage of 25, 50, 100, and 300% treatments. Because most within the Colubridae (Malnate and Underwood 1988, chose the wettest site (300%) for oviposition, we mod- June 2004 FITNESS CONSEQUENCES OF NEST CHOICE 1629 i®ed the protocol to use 100, 300, 400, and 600% treat- wettest substrate (␹2 ϭ 35.41, df ϭ 3, P Ͻ 0.0011). In ments, in order to see whether the snakes simply se- trials providing 100, 300, 400, and 600% water by mass lected the wettest available substrate. We ran 10 further in the alternative nest sites, females again tended to trials with this combination. select the wetter treatments for oviposition (one in Oviposition generally occurred at night, with all eggs 100%, one in 300%, four in 400%, three in 600%) but within a single clutch being laid within a 60-min pe- the effect was not statistically signi®cant (␹2 ϭ 3.0, df riod. Cages were checked each morning and any eggs ϭ 3, P ϭ 0.39). were removed, weighed, and measured, and the eggs To examine whether phenotypic traits of a female from each clutch were separated for incubation under snake or her eggs in¯uenced her choice of incubation a range of hydric conditions. The postpartum females sites, we conducted one-factor ANOVAs with nest hy- were also reweighed at this time, and then released at dric category as the factor and female traits or clutch their site of capture. Eggs from each clutch were al- mean values as the dependent variables. We combined located to four incubation treatments (i.e., split-clutch data from both sets of trials for this analysis. Females design). Eggs were incubated separately in small plas- using the different hydric categories of nests did not tic containers (6.5 cm diameter ϫ 4.5 cm high), partly differ in mean body sizes (SVL, F5,32 ϭ 0.26, P ϭ 0.93; buried in the vermiculite substrate. Each egg was in- mass postpartum, F5,32 ϭ 0.49, P ϭ 0.78), clutch sizes cubated in twice its mass of vermiculite plus an amount (F5,32 ϭ 0.40, P ϭ 0.84), or egg sizes (length, F5,32 ϭ of water determined by its treatment moisture regime. 0.73, P ϭ 0.61; width, F5,32 ϭ 1.13, P ϭ 0.37; mass,

Water was added weekly to replace evaporative loss. F5,32 ϭ 1.87, P ϭ 0.13; see Fig. 1). The incubation containers were kept in a room with Effects of incubation regimes ambient temperatures 24.0ЊϮ8ЊC (mean Ϯ 1 SD; range 16.0±32.0ЊC). Eggs were weighed every week and were on hatchling phenotypes checked daily for hatching; all hatchlings were im- Hatching success was high for eggs in all treatments, mediately measured and weighed. Their muscular regardless of hydric conditions in the nest of origin strength was tested at 1 day of age by attaching a spring (78% of 157 eggs laid in 300% water by mass, 83% balance to the tail and allowing the snake to pull against of 72 eggs laid at drier conditions, and 100% of 28 it seven times in rapid succession; we retained the mean eggs laid in wetter conditions; ␹2 ϭ 3.06, df ϭ 2, P ϭ and maximum strength scores for analysis (see Shine 0.22) or during incubation (84% of 68 eggs incubated and Brown [2002] for details of this method). Hatch- at 300%, 82% of 208 eggs kept in drier conditions, lings were then individually marked by scale-clipping, 80% of 10 eggs kept in wetter conditions; ␹2 ϭ 0.1, df and were released at the site where their mother had ϭ 3, P ϭ 0.95). In this study, eggs incubated only been captured. None of these snakes has yet been re- brie¯y in the initial nest site (from oviposition until captured. However, numerous hatchlings from similar they were removed the following morning, typically a incubation moisture experiments (but only incorporat- few hours) and then spent the rest of incubation (ϳ10 ing 50% and 100% moisture regimes) during the 2000 weeks) at the regime to which they were allocated. and 2001 nesting seasons have been recaptured. Intuition suggests that the initial brief phase will be of trivial importance, but to test this assumption, we in- RESULTS cluded the initial as well as incubation nest conditions in our analysis of hatchling phenotypes. These data Maternal nest-site selection were analyzed with three-factor nested ANOVA, with Of 47 female keelbacks captured shortly prior to egg- the factors being clutch ID number nested within the laying, 38 oviposited within one of the four alternative hydric treatment for the initial nest; the initial nest nest sites provided in the cages. The remaining nine hydric conditions; and the incubation hydric condi- laid their eggs either on the cage ¯oor or in the water tions. Both of these latter factors were tested against bowl; these are excluded from our analyses of the nested term (among-clutch variation) rather than nest-site choice, but their eggs (if viable) were used the residual error term. The analysis also included the for the study of incubation effects (incubated at ex- interaction between hydric conditions in the initial and treme moisture levels of 6, 12, 400, and 600% water, ®nal nests. To simplify the analysis, hydric conditions to bracket the main experimental treatments of 25± were scored as a trichotomous variable (Ͻ300%, 300%, 300% water). and Ͼ300% water). Initial analyses showed that many When female keelbacks were given a choice between of the variables that we measured (tail length, head incubation substrates offering 25, 50, 100, and 300% length, mass, mean and maximum strengths) were high- water by mass, most (21 of 29) selected the wettest ly correlated with SVL (P Ͻ 0.001 in all cases), so we treatment for oviposition. Another four oviposited in calculated residual scores from the general linear re- the driest treatment (25%), three in the 100%, and one gressions of all these traits against SVL. This procedure in the 50%. Contingency-table analysis, against a null generated size-independent measures, so that appar- hypothesis of equal numbers of nests laid in each treat- ently ``signi®cant'' effects on traits were not simple ment, shows that the females signi®cantly preferred the consequences of their correlation with body size. 1630 G. P. BROWN AND R. SHINE Ecology, Vol. 85, No. 6

FIG. 1. Phenotypic traits (mean ϩ 1 SD) of female keelback snakes (Tropidonophis mairii) and their clutches, as a function of the hydric conditions in nest sites in which the captive females laid their eggs. Sample sizes of females in each category were as follows: 25% water, n ϭ 4; 50%, n ϭ 1; 100%, n ϭ 4; 300%, n ϭ 22; 400%, n ϭ 4; 600%, n ϭ 3. See Maternal nest-site selection for statistical tests of these data.

Table 1 shows the main results from these analyses. in the ®rst few hours after the egg was laid, more than Maternal effects were strong for all traits, with signif- two months previously (Table 1). Our analyses did not icant differences among clutches for all of the hatchling reveal any signi®cant interactions between oviposition characteristics that we measured. After allowing for conditions and incubation conditions in these respects this source of variation, some traits also were signi®- (Table 1). Sex ratios of hatchlings were not signi®- cantly affected by the incubation regimes experienced cantly affected by any of our incubation treatments during the egg stage (Fig. 2). Surprisingly, the initial (range 50±63% male, ␹2 ϭ 0.42, df ϭ 2, P ϭ 0.77). nest in which an egg was laid (and remained for only a few hours) had as much in¯uence in this respect as Determinants of survival rates of free-ranging snakes did the incubation regime to which eggs were trans- We released 750 individually marked hatchlings that ferred and kept throughout the remainder (Ͼ99%) of had been incubated on either 50% or 100% moisture development (Table 1). Thus, although hatchling body substrates (including 239 hatchlings from the incuba- size was affected mostly by incubation regime, a ne- tion experiments reported in the current paper) between onate's muscular strength relative to size was strongly July 2000 and September 2002. Hatchlings from wet- in¯uenced by the hydric conditions that it experienced substrate incubation were larger than those from dry- June 2004 FITNESS CONSEQUENCES OF NEST CHOICE 1631

TABLE 1. Three-factor nested ANOVA F values showing the effects of clutch ID number, and hydric conditions in the initial nest and during the subsequent incubation period, on phenotypic traits of hatchling keelbacks (Tropidonophis mairii).

In¯uence of hydric conditions Interaction of initial ϫ incuba- Clutch ID Initial Incubation tion Trait number nest nest conditions Egg mass At laying 71.82*** 1.59 0.02 0.83 At 2 weeks 41.65*** 1.46 0.37 0.16 At 3 weeks 25.49*** 2.70 0.89 0.19 At 4 weeks 18.66*** 3.63* 1.90 0.28 At 5 weeks 13.86*** 3.28 3.50* 0.46 At 6 weeks 8.00*** 3.63* 7.50** 0.43 At 7 weeks 6.82*** 3.27 11.83** 0.59 At 8 weeks 2.61*** 0.56 3.29 0.27 At 9 weeks 6.77*** 0.24 14.54*** 0.03 At 10 weeks 6.13*** 0.50 15.18** 0.89 Snout±vent length (mm) 4.80*** 3.21 18.37*** 0.82 Relative tail length 9.27*** 0.85 0.27 1.14 Relative head length 7.95*** 0.07 0.25 0.83 Relative body mass 5.94*** 0.67 0.44 1.60 Relative mean strength 2.94*** 4.43* 0.76 0.93 Relative maximum strength 2.30** 4.06* 0.99 0.50 Notes: Clutch effects are nested within the initial nest hydric treatment, and the other main effects are tested against this interaction term rather than the residual term. For all variables except snout±vent length, the trait tested is the residual score from the linear regression of the trait against snout±vent length. For clutch ID, df ϭ 26, 197; for tests of hydric conditions (last three columns), all df ϭ 2, 197. * P Ͻ 0.05; ** P Ͻ 0.01; *** P Ͻ 0.001.

substrate incubation (mean values 14.5 vs. 13.3 cm The idea that females select nest sites based upon

SVL, F1,74 ϭ 187.7, P Ͻ 0.0001). In total, 42 of these cues that predict hatchling viability is an obvious one, animals were recaptured 109±1005 days after their re- with a long history (e.g., Fitch 1954, Fitch 1964, Muth lease as hatchlings. The percentage of animals that was 1980). Detailed analyses of several taxa have shown recaptured was higher from wet-substrate incubation that the physical conditions inside natural nests con- (29 of 393, ϭ 7.4%) than from dry-substrate incubation stitute a highly nonrandom sample of the incubation (13 of 357, ϭ 3.6%; ␹2 ϭ 5.94, df ϭ 1, P ϭ 0.026). environments available in potential nest sites (e.g., Logistic regression showed that snakes were more like- Packard and Packard 1988, Shine and Harlow 1996). ly to be recaptured if they were large at hatching (log- Fewer studies, however, have looked at active nest-site likelihood ratio test, ␹2 ϭ 9.68, df ϭ 1, P ϭ 0.002; see selection behavior using appropriate experimental de- Fig. 3). signs to identify the speci®c cues used by nesting fe- males. Such cues include nest temperatures (Bull et al. DISCUSSION 1988) and hydric conditions (Plummer and Snell 1988, Our study shows that female keelbacks actively se- Warner and Andrews 2002). lect moist nest sites; that such nests produce larger Analogous phenomena occur in viviparous reptiles hatchlings; and that larger body size at hatching en- also, with pregnant females maintaining distinctive hances offspring survival. Hence, our data show a di- thermal regimes (often, with low diel variance), and rect link between a female's nest-site choice and her thus controlling the physical conditions under which reproductive success, mediated via phenotypically their offspring develop (Beuchat 1986, Charland and plastic responses during embryogenesis. The hypoth- Gregory 1990, Gregory et al. 1999). This behavior may esis that reproducing females enhance their own ®tness substantially affect the phenotypic traits of offspring by manipulating the phenotypic traits of their offspring (Beuchat 1988, Shine and Harlow 1993, Swain and (via nest-site selection) is consistent with indirect ev- Jones 2000, Wapstra 2000, Arnold and Peterson 2002). idence from a variety of previous studies. However, Active maternal control over the incubation environ- our study is unusual in providing direct evidence for ment thus is widespread among reptiles. Although ther- each of the major links in the hypothesis, rather than mal factors have been the primary focus of previous relying upon assumptions about critical aspects such studies, this probably re¯ects the concentration of re- as maternal criteria for nest choice, and the phenotypic search on temperate-zone (cool-climate) reptile taxa. determinants of hatchling ®tness. In many tropical areas, high temperatures are available 1632 G. P. BROWN AND R. SHINE Ecology, Vol. 85, No. 6

FIG. 2. Phenotypic traits (mean ϩ 1 SD) of hatchling keelbacks (Tropidonophis mairii) as a function of (a, b) the hydric conditions in nest sites in which the captive females laid their eggs and (c, d) the hydric regimes under which eggs were kept during the 10-week incubation period. Sample sizes (number of eggs) were as follows: for initial treatment, 25% water, n ϭ 37; 50%, n ϭ 10; 100%, n ϭ 25; 300%, n ϭ 163; 400%, n ϭ 28; 600%, n ϭ 15; for subsequent incubation treatment, 6%, n ϭ 5; 12%, n ϭ 5; 25%, n ϭ 72; 50%, n ϭ 66; 100%, n ϭ 60; 300%, n ϭ 68; 400%, n ϭ 5; 600%, n ϭ 5. See Table 1 for statistical tests of these data. year-round and the most important seasonal variation Shine 1999). Studies on squamate reptiles have typi- for incubating eggs will be hydric, rather than thermal, cally focused on thermal, rather than hydric, in¯uences, in nature. with the latter found to exert little or no effect on em- We obtained strong support for the hypothesis that bryogenic pathways in some taxa (e.g., Ji and Brana female keelbacks base nest-site choice at least partly 1999, Flatt et al. 2001). However, many turtles show on moisture content of the substrate, but saw no hint strong responses to minor variation in water potential that females match their choice of oviposition site to of the incubation medium (Packard 1991, Packard et the reaction norms of their own offspring. One could al. 1993), and the same is clearly true for some squa- imagine a situation whereby some females within the mate taxa also, including Tropidonophis (Shine and population specialize in dry-substrate incubation (i.e., Brown 2002). Drier substrates result in hatchlings that produce eggs that bene®t from incubation on drier sub- are small because they are unable to resorb all of the strates, and actively select such substrates), whereas yolk sac and thus leave this behind in the egg (Shine other females specialize on moist substrates. If present, and Brown 2002). The most surprising result from our such an effect should be revealed by signi®cant inter- current analyses is that the water content of the incu- actions between nest-site choice and incubation re- bation medium in which an egg is laid somehow exerts sponses, but no such interactions were apparent in our a signi®cant in¯uence on the muscular strength of the analyses (Table 1). A study of montane reached neonatal snake that emerges from that egg months later, the same conclusion, based on ®eld data (Shine et al. even if the initial exposure is very brief (a few hours, 1997). out of a 10-week incubation period). The proximate Phenotypic plasticity in response to physical con- basis for this effect may involve changes to the eggshell ditions during egg incubation is widespread among rep- or embryonic membranes as they take up or lose water tiles, from a diversity of phylogenetic lineages and hab- immediately after oviposition, and warrants further itat types (see Deeming and Ferguson 1991; review by study. June 2004 FITNESS CONSEQUENCES OF NEST CHOICE 1633

ACKNOWLEDGMENTS We thank T. Tam and M. DaQuieri for discussions, the staff at Beatrice Hill Farm for logistical support, and the Australian Research Council for funding.

LITERATURE CITED Albon, S. D., F. E. Guinness, and T. H. Clutton-Brock. 1983. The in¯uence of climatic variation on the birth weights of red deer (Cervus elaphus). Journal of Zoology (London) 200:295±297. Arnold, S. J., and A. Bennett. 1988. Behavioural variation in natural populations. V. Morphological correlates of lo- comotion in the garter snake Thamnophis radix. Biological Journal of the Linnean Society 34:175±190. Arnold, S. J., and C. R. Peterson. 2002. A model for optimal FIG. 3. Survival (recapture) rates of free-ranging keel- reaction norms: the case of the pregnant garter snake and backs (Tropidonophis mairii) as a function of their body size her temperature-sensitive embryos. American Naturalist at hatching. The histograms show the percentage of recap- 160:306±316. tured snakes of each 1-cm snout±vent length (SVL) class, Beuchat, C. A. 1986. Reproductive in¯uences on the ther- except that SVL categories are combined where N Ͻ 10. Thus, moregulatory behaviour of a live-bearing . Copeia the 11-cm class (N ϭ 71) contains 14 offspring Ͻ11 cm, and 1986:971±979. the 16-cm class (N ϭ 28) contains three offspring Ͼ16 cm. Beuchat, C. A. 1988. Temperature effects during gestation Sample sizes for the other size classes are as follows: 107 in a viviparous lizard. Journal of Thermal Biology 13:135± for 12 cm; 144 for 13 cm; 263 for 14 cm, and 137 for 15 cm. 142. The dotted line shows the probability equation from the lo- Beuchat, C. A., and S. Ellner. 1987. A quantitative test of gistic regression of survival against body size at hatching. life history theory: thermoregulation by a viviparous lizard. Ecological Monographs 57:45±60. Brown, G. P., and R. Shine. 2002. Reproductive ecology of a tropical natricine snake, Tropidonophis mairii (Colubri- dae). Journal of Zoology (London) 258:63±72. The ®nding that larger body size at hatching en- Brown, G. P., R. Shine, and T. Madsen. 2002. Responses of hances the probability of survival for young keelbacks three sympatric snake species to tropical seasonality in accords well with intuition, and with results from sev- northern Australia. Journal of Tropical Ecology 18:549± 568. eral mark±recapture studies on other species of reptiles Bull, J. J. 1983. Evolution of sex determining mechanisms. (e.g., Fox 1975, Ferguson and Fox 1984). Nonetheless, Benjamin/Cummings Publishing, Menlo Park, California, many other studies have failed to ®nd any evidence USA. that ``bigger is better,'' including a detailed analysis of Bull, J. J. 1987. Evolution of phenotypic variance. Evolution 41:303±315. survival rates in water pythons (Liasis fuscus) at the Bull, J. J., W. H. N. Gutzke, and M. G. Bulmer. 1988. Nest same study areas that we use for our keelback studies choice in a captive lizard with temperature-dependent sex (Madsen and Shine 1998). The role of hatchling size determination. Journal of Evolutionary Biology 2:177±184. in predicting survival may also vary among years even Bull, J. J., R. C. Vogt, and M. G. Bulmer. 1982. Heritability of sex ratio in turtles with environmental sex determination. within a single population (Fox 1975, Sinervo and Evolution 36:333±341. Huey 1990). This variability makes it dif®cult to infer Burger, J., R. T. Zappalorti, and M. Gochfeld. 1987. Devel- the direction and magnitude of selective forces on off- opmental effects of incubation temperature on hatchling spring traits without very long-term data. In the case pine snakes Pituophis melanoleucus. Comparative Bio- chemistry and Physiology 87A:727±732. of Tropidonophis at Fogg Dam, larger hatchlings have Charland, M. B., and P. T. Gregory. 1990. The in¯uence of survived signi®cantly better than smaller conspeci®cs female reproductive status on thermoregulation in a vivip- in each of the three years for which we have recapture arous snake, Crotalus viridis. Copeia 1990:1089±1098. Cogger, H. G. 2000. Reptiles and of Australia data (G. P. Brown and R. Shine, unpublished data). Sixth edition. Reed New Holland, Sydney, New South In summary, our results are encouraging for those Wales, Australia. who infer an adaptive signi®cance to maternal nest site Deeming, D. C., and M. W. J. Ferguson. 1991. Physiological choice. Female keelbacks actively select among alter- effects of incubation temperature on embryonic develop- ment in reptiles and birds. Pages 147±171 in D. C. Deeming native nest sites based on a trait (substrate water po- and M. W. J. Ferguson, editors. Egg incubation: its effects tential) that in¯uences the body size of hatchlings, on embryonic development in birds and reptiles. Cam- which in turn affects the survival of these offspring. bridge University Press, Cambridge, UK. The direction of these effects conforms to the adap- Ferguson, G. W., and S. F. Fox. 1984. Annual variation of survival advantage of large juvenile side-blotched lizards, tationist prediction: that is, the females' preference for Uta stansburiana: its causes and evolutionary signi®cance. moister sites enhances, rather than reduces, offspring Evolution 38:342±349. ®tness. By using a cue that directly predicts develop- Fitch, A. V. 1964. Temperature tolerances of embryonic Eu- mental responses of their embryos, female keelbacks meces. Herpetologica 20:184±187. Fitch, H. S. 1954. Life-history and ecology of the ®ve-lined can thereby maximize the ®tness bene®ts from egg pro- skink, Eumeces fasciatus. University of Kansas, Publica- duction. tions of the Museum of Natural History 8:1±156. 1634 G. P. BROWN AND R. SHINE Ecology, Vol. 85, No. 6

Flatt, T., R. Shine, P. A. Borges-Landaez, and S. J. Downes. Rhen, T., and J. W. Lang. 1995. Phenotypic plasticity for 2001. Phenotypic variation in an oviparous montane lizard growth in the common snapping turtle: effects of incubation (Bassiana duperreyi): the effects of thermal and hydric in- temperature, clutch, and their interaction. American Nat- cubation environments. Biological Journal of the Linnean uralist 146:726±747. Society 74:339±350. Scheiner, S. M., and H. S. Callahan. 1999. Measuring natural Fox, S. F. 1975. Natural selection on morphological pheno- selection of phenotypic plasticity. Evolution 53:1704± types of the lizard Uta stansburiana. Evolution 29:95±107. 1713. Gregory, P. T., L. H. Crampton, and K. M. Skebo. 1999. Shine, R. 1991. Strangers in a strange land: ecology of the Con¯icts and interactions among reproduction, thermoreg- Australian colubrid snakes. Copeia 1991:120±131. ulation and feeding in viviparous reptiles: are gravid snakes Shine, R. 1999. Why is sex determined by nest temperature anorexic? Journal of Zoology 248:231±241. in many reptiles? Trends in Ecology and Evolution 14:186± 189. Henry, C. J. K., and S. J. Ulijaszek. 1996. Long-term con- Shine, R., and G. P. Brown. 2002. Effects of seasonally vary- sequences of early environment. Cambridge University ing hydric conditions on hatchling phenotypes of keelback Press, Cambridge, UK. snakes (Tropidonophis mairii, Colubridae) from the Aus- Jayne, B. C., and A. Bennett. 1990. Selection on locomotor tralian wet-dry tropics. Biological Journal of the Linnean performance capacity in a natural population of garter Society 76:339±347. snakes. Evolution 44:1204±1229. Shine, R., M. J. Elphick, and P. S. Harlow. 1997. The in¯u- Ji, X., and F.Brana. 1999. The in¯uence of thermal and hydric ence of natural incubation environments on the phenotypic environments on embryonic use of energy and nutrients, traits of hatchling lizards. Ecology 78:2559±2568. and hatchling traits, in the wall lizards (Podarcis muralis). Shine, R., and P. Harlow. 1993. Maternal thermoregulation Comparative Biochemistry and Physiology A 124:205± in¯uences offspring viability in a viviparous lizard. Oec- 213. ologia 96:122±127. Joanen, T., L. McNease, and M. W. J. Ferguson. 1987. The Shine, R., and P. S. Harlow. 1996. Maternal manipulation of effects of egg incubation temperature on post-hatching offspring phenotypes via nest-site selection in an oviparous growth of American alligators. Pages 533±537 in G. J. W. reptile. Ecology 77:1808±1817. Webb, S. C. Manolis, and P. J. Whitehead, editors. Wildlife Sinervo, B., and R. B. Huey. 1990. Evolution of body size management: crocodiles and alligators. Surrey Beatty and in hatchling lizardsÐtesting Lack's hypothesis. American Sons, Sydney, New South Wales, Australia. Zoologist 30:A56±A56. Madsen, T., and R. Shine. 1998. Quantity or quality? Deter- Sultan, S. 1987. Evolutionary implications of phenotypic minants of maternal reproductive success in tropical py- plasticity in plants. Evolutionary Biology 21:127±178. thons (Liasis fuscus). Proceedings of the Royal Society of Swain, R., and S. M. Jones. 2000. Maternal effects associated London, Series B 265:1521±1525. with gestation conditions in a viviparous lizard, Niveos- cincus metallicus. Herpetological Monographs 14:432± Malnate, E. V., and G. Underwood. 1988. Australasian na- 440. tricine snakes of the Tropidonophis. Proceedings of Taning, A. V. 1952. Experimental study of meristic characters the Academy of Natural Sciences of Philadelphia 140:59± in ®shes. Biological Review 27:169±193. 201. Via, S., R. Gomulkiewicz, G. De Jong, S. M. Scheiner, C. D. Muth, A. 1980. Physiological ecology of desert iguana (Dip- Schlichting, and P. H. Van Tienderen. 1995. Adaptive phe- sosaurus dorsalis) eggs: temperature and water relations. notypic plasticity: consensus and controversy. Trends in Ecology 61:1335±1343. Ecology and Evolution 10:212±217. O'Shea, M. 1991. The reptiles of Papua New Guinea. British Wapstra, E. 2000. Maternal basking opportunity affects ju- Herpetological Society Bulletin 37:15±32. venile phenotype in a viviparous lizard. Functional Ecology Packard, G. C. 1991. Physiological and ecological impor- 14:345±352. tance of water to embryos of oviparous reptiles. Pages 213± Warkentin, K. M. 1995. Adaptive plasticity in hatching ageÐ 228 in D. C. Deeming and M. W. J. Ferguson, editors. Egg a response to predation risk trade-offs. Proceedings of the incubation: its effects on embryonic development in birds National Academy of Sciences (USA) 92:3507±3510. and reptiles. Cambridge University Press, Cambridge, UK. Warner, D. A., and R. M. Andrews. 2000. Maternal and en- Packard, G. C., K. Miller, and M. J. Packard. 1993. Envi- vironmental contributions to hatchling phenotypes and sur- ronmentally induced variation in body size of turtles hatch- vival of the lizard Sceloporus undulatus. American Zool- ogist 40:1251±1251. ing in natural nests. Oecologia 93:445±448. Warner, D. A., and R. M. Andrews. 2002. Nest-site selection Packard, G. C., and M. J. Packard. 1988. The physiological in relation to temperature and moisture by the lizard Sce- ecology of reptilian eggs and embryos. Pages 523±605 in loporus undulatus. Herpetologica 58:399±407. C. Gans and R. B. Huey, editors. Biology of the Reptilia. Webb, J. K., G. P. Brown, and R. Shine. 2001. Body size, Volume 16. Alan R. Liss, New York, New York, USA. locomotor speed and antipredator behaviour in a tropical Plummer, M. V., and H. L. Snell. 1988. Nest site selection snake (Tropidonophis mairii, Colubridae): the in¯uence of and water relations of eggs in the snake Opheodrys aestivus. incubation environments and genetic factors. Functional Copiea 1988:58±64. Ecology 15:561±568.